Clements M.,Signetron Inc. |
Zakhor A.,Signetron Inc.
2014 IEEE International Conference on Image Processing, ICIP 2014 | Year: 2014
Image geo-localization is an important problem with many applications such as augmented reality and navigation. The most common ways to geo-localize an image are to use its meta-data such as GPS or to match it against a geotagged database. When neither of those is available, it is still possible to apply shadow analysis to determine the camera heading for outdoor images. This could be useful pruning the search space in geo-localization applications, for example by removing roads with incompatible orientations from a database such as Open Street Map. In this paper, we develop a novel interactive method for deducing the global heading of a query image using the shadows in it. We start by constructing a model of the sun-earth system to determine all shadows possible at a given approximate latitude, and compare shadows within the query to those possible under the model to determine the range of possible headings. We demonstrate this on 54 query images with known ground truth, and show that in 52 cases the ground truth lies in the computed range. © 2014 IEEE. Source
Singh G.,Signetron Inc. |
Singh G.,University of California at Berkeley |
Jouppi M.,Signetron Inc. |
Jouppi M.,University of California at Berkeley |
And 4 more authors.
Proceedings of SPIE - The International Society for Optical Engineering | Year: 2015
Automatic building extraction in satellite imagery is an important problem. Existing approaches typically involve stereo processing two or more satellite views of the same region. In this paper, we use shadow analysis coupled with line segment detection and texture segmentation to construct rectangular building approximations from a single satellite image. In addition, we extract building heights to construct a rectilinear height profile for a single region. We characterize the performance of the system in rural and urban regions of Jordan, Philippines, and Australia and demonstrate a detection rate of 76.2 - 86.1% and a false alarm rate of 26.5 - 40.1%. © 2015 SPIE-IS & T. Source